2018 vs. 2019 QB class

[Gmen]

I've been getting into analytics lately and I decided to compare the 2018 QB class to the 2019 QB class using the two most popular QB analytics metrics : expected points added (EPA) and completion percentage over expectation (CPOE).  Here are the results.

 

 

 

[SDotNova]

Nice chart. For most of the 2018 QBs, this year is an important year. Right now based on that chart, all we can say is that Rosen is likely a bust. All these other guys are jumbled up. Trend wise, it kind of fits what I expected: Lamar significantly improved, Darnold and Allen improved and Baker regressed. Kyle Allen 👀. Should be fun to watch who comes out of Washington as the guy.

Edited May 11, 2020 by SDotNova

[Gmen]

20 minutes ago, SDotNova said:

Nice chart. For most of the 2018 QBs, this year is an important year. Right now based on that chart, all we can say is that Rosen is likely a bust. All these other guys are jumbled up. Trend wise, it kind of fits what I expected: Lamar significantly improved, Darnold and Allen improved and Baker regressed. Kyle Allen 👀. Should be fun to watch who comes out of Washington as the guy.

Thanks.  It's worth noting that this CPOE model was made without tracking data and it just models the CPOE using simple statistics like throw distance, down, yards to go.  However, Next Gen Stats has a model developed with player tracking data, so they can factor in information like how close a defender is the receiver, how close a defender is to the quarterback, etc.  I put a lot more stock in the Next Gen model because it has access to more advanced data. That model is much more favorable to Daniel Jones than the nflfastR model.  His CPOE is -0.2% as opposed to -2.6% with this model.

https://nextgenstats.nfl.com/stats/passing#completionPercentageAboveExpectation

And here is a write-up about how they developed their model

http://www.nfl.com/news/story/0ap3000000964655/article/next-gen-stats-introduction-to-completion-probability

[Danger]

2018, for now.